Li et al. (2025) PAGrid: A probabilistic area-weighted gridding method for seamless mapping of sentinel-3 swath data

Identification

• Journal: Remote Sensing of Environment
• Year: 2025
• Date: 2025-12-01
• Authors: Dong Li, Anirudh Belwalkar, Tao Cheng, Kang Yu
• DOI: 10.1016/j.rse.2025.115165

Research Groups

• Precision Agriculture Laboratory, School of Life Sciences, Technical University of Munich, Munich, Germany
• National Engineering and Technology Center for Information Agriculture (NETCIA), Nanjing Agricultural University, Nanjing, China

Short Summary

This study proposes a probabilistic area-weighted gridding method (PAGrid) to overcome spatial discontinuities in gridded Sentinel-3 OLCI swath data. PAGrid efficiently generates more consistent and continuous gridded time series by approximating area-weighting through randomized spatial perturbations, significantly outperforming conventional center point-based gridding.

Objective

• To evaluate the proposed Probabilistic Area-Weighted Gridding (PAGrid) method and the traditional Center Point-based Gridding (CPGrid) method in mapping Sentinel-3 Ocean and Land Colour Instrument (OLCI) data.

Study Configuration

• Spatial Scale: Germany and global scale; nominal grid spacing of 300 meters (1/336° latitude-longitude grid); OLCI ground sampling distance of 300 meters at nadir, varying up to 1 kilometer across-track.
• Temporal Scale: Entire year of 2022; daily observations aggregated into 8-day composites.

Methodology and Data

• Models used:
  • Probabilistic Area-Weighted Gridding (PAGrid)
  • Center Point-based Gridding (CPGrid)
  • Area-Weighted Gridding (AGrid) (mentioned but not applied due to computational cost)
• Data sources:
  • Sentinel-3A and Sentinel-3B Ocean and Land Colour Instrument (OLCI) Level-2 surface reflectance swath data (Product IDs: S3ASY2SYN and S3BSY2SYN) for 2022.
  • Canopy absorption coefficient by chlorophyll in the red-edge band (αRE) derived from OLCI reflectance bands (R709 and R865).

Main Results

• PAGrid increased the median percentage of valid grid cells from 85% (CPGrid) to 93%.
• PAGrid reduced temporal fluctuations in the seasonal αRE series by 21% compared to CPGrid (average Standard Deviation of First-order Difference (SFD) of 0.26 for PAGrid vs. 0.33 for CPGrid).
• PAGrid improved the mean correlation (R²) between Sentinel-3A and Sentinel-3B αRE from 0.73 (CPGrid) to 0.84.
• PAGrid demonstrated enhanced spatial continuity and reduced spatial gaps in gridded αRE maps, particularly at high viewing zenith angles.
• At a global scale, PAGrid produced a higher fraction of valid grid cells and improved inter-sensor consistency (R² from 0.93 to 0.96 for 8-day composites) compared to CPGrid.
• PAGrid is computationally efficient for large-scale applications, being roughly two orders of magnitude slower than CPGrid (23.17 seconds vs. 0.18 seconds for Germany with 100 iterations) but significantly faster than the exact AGrid method (23.17 seconds vs. 29,816 seconds).

Contributions

• Introduces PAGrid, a novel probabilistic area-weighted gridding method that balances accuracy and computational efficiency for large-scale satellite imagery.
• Circumvents complex geometric calculations of traditional area-weighted methods by simulating overlap probability through randomized spatial perturbations.
• Generates more consistent and continuous gridded time series from swath-based satellite observations, specifically demonstrated with Sentinel-3 OLCI data.
• Enhances spatial completeness and reduces temporal fluctuations in gridded remote sensing products.
• Offers a generalizable framework applicable to other swath-based sensors (e.g., MODIS, VIIRS), improving their usability for time series analysis and environmental monitoring.

Funding

• LiveSen-MAP project funded by the European Commission
• National Natural Science Foundation of China (Grant No. 42101360)

Citation

@article{Li2025PAGrid,
  author = {Li, Dong and Belwalkar, Anirudh and Cheng, Tao and Yu, Kang},
  title = {PAGrid: A probabilistic area-weighted gridding method for seamless mapping of sentinel-3 swath data},
  journal = {Remote Sensing of Environment},
  year = {2025},
  doi = {10.1016/j.rse.2025.115165},
  url = {https://doi.org/10.1016/j.rse.2025.115165}
}